A geometric approach for the analysis and computation of the intrinsic camera parameters

The authors of this paper adopted the projected characteristics of the abso lute conic in terms of the Pascal's theorem to propose an entirely new came ra calibration method based on purely geometric thoughts. The use of this t heorem in the geometric algebra framework allows us to compute a projective invariant using the conics of only two images which expressed using bracke ts helps us to set enough equations to solve the calibration problem. The m ethod requires restricted controlled camera movements. Our method is less s ensitive to noise as the Kruppa's-equation-based methods. Experiments with simulated and real images confirm that the performance of the algorithm is reliable. (C) 2001 Pattern Recognition Society. Published by Elsevier Scien ce Ltd. All rights reserved.